US agencies accuse six Chinese AI firms of distilling frontier models and recommend new defenses that could affect enterprise AI access and API use.